The number of substance abuse program admissions for youths in the United States has been rising dramatically, increasing by 46 percent from 1993 to 1998. Despite this growth, little is known about the effectiveness of community-based treatment services for adolescents. Previous studies find that some adolescent outcomes improve after treatment and that outcomes vary across service modalities. These studies, however, are descriptive. They do not examine whether the services youths receive cause the observed functional improvements, or whether these improvements instead are caused by factors unrelated to treatment services, like maturation, the natural course of adolescent psychosocial problems or pretreatment differences in groups. Determining the causal effect attributable to adolescent services is more complex, but policy relevant: unless we know the relative improvement attributable to a particular form of treatment, time-in-treatment or source of referral, we can neither evaluate the relative cost-effectiveness of alternative services nor determine the value in funding such services. The proposed study estimates the causal effects of community-based adolescent services using data collected in NIDA's Drug Abuse Treatment Outcomes Studies for Adolescents (DATOS-A). Innovative statistical methods are planned to distinguish pretreatment group differences from differential treatment effects in the explanation of 12-month post treatment outcomes of youths receiving different service modalities. The study has four specific aims: 1) estimate the causal effects of outpatient, residential and short-term inpatient services on the outcomes of those adolescents most likely to enter each program; 2) for each modality, estimate the causal effects of treatment length; 3) compare the effectiveness of court-referred services to those provided without legal pressure; and 4) develop and evaluate statistical methods required for aims 1 to 3, and compare findings using our new methods to those produced using methods more standard in substance abuse treatment services research. This study will provide the most rigorous and conclusive assessment of the comparative effectiveness of community-based adolescent substance abuse programs yet conducted. In addition, it will develop and disseminate more relevant and robust causal modeling approaches to substance abuse treatment services research than are currently used in our field.